Gene expression profiling of peripheral blood from athletes administered growth hormone (GH) for 8 weeks found:
1) GH increased circulating insulin-like growth factor I (IGF-I) levels approximately 2-fold in both men and women.
2) Microarray analysis detected small changes in gene expression with GH, with a maximum 2-fold increase or decrease. 353 genes were differentially expressed in women and 41 in men.
3) The effect of GH on gene expression was similar in magnitude to natural variation between individuals, making it an unlikely approach for detecting GH doping.
2. that may undergo characteristic changes in gene expres-
sion in response to treatment. We hypothesized that GH
may induce effects on the peripheral blood transcriptome,
which could be used as the basis for a GH doping test. The
aim of this study is to determine whether a simple test for
GH doping can be through the analysis of peripheral
blood transcriptomes in humans.
Subjects and Methods
Study design
This study is drawn from a larger project analyzing the phar-
macodynamics of markers of GH abuse in serum, the results of
which have been recently published (6). Briefly, this was a double-
blind, placebo-controlled study in male and female recreational
athletes comprising an 8-wk treatment phase followed by 6 wk
washout. Average ages of participants were 24.3 Ϯ 2.4 yr
(mean Ϯ SE) for men and 31.7 Ϯ 1.6 yr for women. This study
comprised seven men and 13 women treated with GH 2 mg/d sc
in whom peripheral blood mononuclear cells were obtained for
gene expression analysis. GH (1 mg/ml Norditropin; Novo Nor-
disk, Bagsvaerd, Denmark) was self-administered at night. To
minimize side effects, the dose was increased from 1 mg/d (first
week) to 1.5 mg/d (second week) and then to the final dose of 2
mg/d for 6 wk. Participants attended the Clinical Research Fa-
cility at the end of each week, when the injection cartridge was
changed and compliance checked by verbal reports and the vol-
ume remaining in the cartridge. Blood was collected at baseline
(wk 0), during and after the final treatment point (wk 4 and 8),
and after 6 wk washout (wk 14). RNA was prepared for mi-
croarray analysis at baseline and after the final GH treatment.
The study was approved by the Human Research Ethics Com-
mittee, St. Vincent’s Hospital.
Determination of serum IGF-I concentration
Serumsampleswerestoredat–80Cbeforebatchanalysis.All
samples for any individual were measured in the same assay run.
IGF-I concentration (intraassay and interassay coefficients of
variation of Ͻ4 and Ͻ9%, respectively) was measured by RIA
after acid-ethanol extraction (13), using iodinated des(1-3)IGF-I
(GroPep, Adelaide, South Australia) as radioligand.
RNA collection
RNA was extracted from white blood cells at the time of
blood collection to avoid degradation within 6 h. Blood (14 ml)
was collected by venipuncture, and after red cell lysis, RNA was
extracted from the washed white cells using Trizol (Invitrogen,
Waverly, Australia) with yields of 25–80 ng and stored under
ethanolatϪ80Cbeforemicroarrayanalysis.Gelelectrophoresis
was performed to assess the integrity of each RNA sample. Be-
foremicroarrayanalysis,RNAwaspurifiedbyRNeasyminispin
columns (QIAGEN, Doncaster, Australia), and the concentra-
tion and quality was assessed by NanoDrop spectrophotometer
and Agilent bioanalyzer, respectively.
Microarray and data analysis
Microarray analysis using the Agilent microarray platform
was performed at The Ramaciotti Centre for Gene Function
Analysis, University of New South Wales, Sydney. The Agilent
Technologies (Santa Clara, CA) platform comprises a two-color
design with two RNA samples labeled with either Cy3 or Cy5
and hybridized to the same 44K G4112F Agilent microarray.
Each array contains about 41,000 unique noncontrol 60-mer
probes. A total of 40 arrays on 10 slides were performed. Base-
line (wk 0) and final GH treatment (wk 8) samples from seven
men and 13 women were two-color labeled and hybridized to
randomly chosen arrays. A dye swap was performed for each
individual on a different slide to remove any bias from the
labeling dyes.
To obtain the relative expression level of the two samples
hybridized on each array, we used the Agilent log ratio (ALR),
which is the log ratio of the processed red/green channel for each
probe. ALRs were transformed to log base 2 and then scale nor-
malized. After correcting for the dye swap, the following linear
model was fitted to the data (14) from each probe independently:
yi ϭ ␣i ϩ i ϩ i, where yi are the normalized expression ratios,
␣i and i are the average ratios for the male and female samples,
respectively, and i is the residual error for probe i. To account
for the additional correlation that is expected from dye-swapped
technical replicates, we estimated the duplicate correlation
among samples (15) and used the sample ID as a blocking vari-
able in the linear model fit. To assess the extent of differential
expression, an empirical Bayes, moderated t statistic (14) was
performed. All analyses were performed using R version 2.7.1
(16) and the Limma package version 2.14 (14) from Bioconduc-
tor version 2.2 (17).
Microarray data are presented on a log2 scale where each unit
represents a 2-fold change in expression level. The absolute fold
change is used to denote changes in expression; i.e. a 2-fold in-
crease is equivalent to a doubling of expression and a 2-fold
decrease a halving of expression. For analysis of differentially
regulated probes, we included only those with a log2 expression
level of 6.0 or greater, signifying good expression above back-
ground. For comparison, the log2 expression for GAPDH was
approximately 13.8.
For probe effect size comparison, we obtained the average
expression level (A value) for each probe: A ϭ (R ϩ G)/2, where
R and G are the background-corrected, log2-transformed mean
intensities for the red and green channels, respectively. The av-
erage expression ratio (M value) for each probe was the ALR as
before, where M ϭ R Ϫ G. Thus, from M and A values, we
obtained the normalized expression level for each channel R and
G for each probe. Duplicate measurements were averaged to
obtain expression level of each subject at baseline and wk 8. The
variance of each probe attributable to GH was determined by
averaging the variance of the two measurements for each indi-
vidual. The variance of each probe between individuals was de-
termined by averaging the expression level for each individual.
Datafromindividualchannelswerealsousedformultivariate
analysis using stepwise diagonal discriminant analysis (SDDA)
an algorithm developed by the Commonwealth Scientific and
Industrial Research Organization. The method is an extension of
diagonal discriminant analysis (18) that uses a forward stepwise
approach to identify a discriminatory set of probes. It identifies
linear combinations of probes that predict the samples from
treatment time points from basal samples. The discriminatory
power of individual probes or groups of probes identified was
estimated by a cross-validation process. Under cross-validation,
individual samples are removed and their status predicted using
4704 Mitchell et al. GH Doping Detection by Blood Microarray J Clin Endocrinol Metab, December 2009, 94(12):4703–4709
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.
3. SDDAappliedtotheremainingsamplesandtheprocessrepeated
leaving out each sample (or individual) in turn.
Statistics
Where presented, P values are unadjusted for multiple testing
bythepositivefalsediscoveryrate(FDR)(19).SerumIGF-Ilevels
and relative expression levels for microarray and quantitative
PCR (Q-PCR) analysis are presented as the mean Ϯ SE.
Q-PCR
RNA samples for seven men and 13 women were used for
PCR analysis. cDNA was synthesized from 2 g total RNA, 25
g/ml oligo deoxy-thymidine (dT), 2.5 M random primers (Pro-
mega, Sydney, Australia) using Superscript II (Invitrogen, Carls-
bad, CA) following the manufacturer’s instructions. Reactions
were diluted 1:4 and stored at Ϫ80 C before Q-PCR. Validation
of the selected genes was undertaken via TaqMan gene expres-
sion assays using the Applied Biosystems 7900HT (Applied Bio-
systems, Foster City, CA). For each gene, an assay was selected
to amplify the region corresponding to the location of the rele-
vant probe. Where a corresponding probe was not available, an
alternative validated probe was selected for the gene. The house-
keepinggeneGAPDHwasperformedinparallelascontrol.Stan-
dard TaqMan cycling conditions were used, and all reactions
were performed in triplicate. Changes in expression level by Q-
PCR were calculated as Ϫ⌬⌬CT (cycle threshold) and presented
relative to average changes derived from ALRs as mean Ϯ SE, on
a log2 scale of relative expression (20). Increased expression has
a positive value, no change zero, and decreased expression a
negative value.
Genes were selected for Q-PCR validation from the top 200
probes. Targets were selected if more than one probe corre-
spondingtoasinglegenewassignificantlyregulated.Anyfurther
probes for that gene that was not in the original selection were
also considered. Each individual probe was aligned against the
genome to confirm selective targeting of an individual transcript.
Results
In both men and women, GH treatment induced a ro-
bust increase in serum IGF-I. The mean IGF-I concen-
tration was elevated at wk 4 and remained elevated at
wk 8, returning to baseline at wk 14 (Fig. 1). IGF-I levels
in men and women increased 2.3- and 1.8-fold, respec-
tively, at 8 wk.
Gene expression changes
For both men and women, most probes cluster around
zero in the log2 scale, indicating no significant change in
expression (Fig. 2, A and D). The maximal ALR changes
in both men and women for individual probes are approx-
imately a doubling or halving. The distribution of P values
was similar in men and women (Fig. 2, B and E) with 11.4
and 10.2% of all detected probes having P Ͻ 0.05, re-
spectively. The relationship between P value and log fold
change for all probes in men and women are summarized
in Fig. 2, C and F. The horizontal dashed line indicates the
P value threshold corresponding to a FDR of 5%, where
there are 41 and 353 probes differentially regulated in men
and women, respectively. The different numbers of arrays
in men and women led to differences in the statistical
power and, thus, the threshold for false discovery. There
were seven and five differentially regulated probes with
bothanALRmorethanϮ0.585,correspondingtoaϮ1.5-
fold change at a FDR less than 0.05, in men and women,
respectively(opencirclesinFig.2,CandF).Thus,ALRfor
all probes revealed that the effect of GH was small.
Results from individual Cy3- and Cy5-labeled channels
were used to estimate the inter-individual and GH effects
across all probes. Analysis of either labeled channel
yielded similar results. An estimation of the relative effects
can be represented by plotting the within- and between-
sample variances as a box plot (Fig. 3). For both men and
women, the most extreme values are inter-individual ef-
fects, indicating that the largest effects are due to the dif-
ference in expression levels between individuals. In men,
the median (0.103) and the range of the middle 50% of the
data (0.059–0.192) for a GH effect was greater than the
inter-individual effect, median 0.070 and range 0.038–
0.134. However, the effect of GH remains similar to the
variation between individuals. In women, the median was
lower (0.077), and the range of the middle 50% of the
variance (0.044–0.203) overlapped the range for men.
Inter-individual variance in women was similar to men,
median 0.078 and range 0.44–0.140. In brief, the effect of
GH is similar to the magnitude of the variation between
individuals.
Gender comparison
The maximal fold changes in expression induced by
GH were similar in men and women. In men, this ranged
from 2.09-fold increase to 1.96-fold decrease and in
women from 2.02-fold increase to 1.82-fold decrease. In
men, there were 36 probes with greater than 1.5-fold up-
regulation and 13 with a similar degree of down-regula-
0
50
100
150
200
250
300
350
400
0 2 4 6 8 10 12 14
Time (Weeks)
Men
Women
FIG. 1. GH-induced changes in IGF-I levels in microarray analysis
subjects. IGF-I was measured at baseline (wk 0), during and after the
final GH treatment (wk 4 and 8), and after 6 wk washout (wk 14).
Results are shown as micrograms per liter IGF-I in serum in seven
GH-treated men and 13 women, mean Ϯ SE.
J Clin Endocrinol Metab, December 2009, 94(12):4703–4709 jcem.endojournals.org 4705
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.
4. tion, representing about 0.1% of all probes. For women,
only 16 probes were regulated by more than 1.5-fold with
eight up-regulated and eight down-regulated (Table 1).
Approximately 11% of all detected probes were differen-
tially regulated (P Ͻ 0.05 and log2 expression Ն6.0) in
men and women (Table 1). In men, the majority, 1795,
was up-regulated compared with 1336 that were down-
regulated.Inwomen,theresultsweresimilar;2799probes
were differentially regulated, with 1465 up-regulated and
1334 down-regulated. More probes were identified in
women at all cutoff values. At an FDR of 5%, 41 probes
were identified in men, 28 (68%) up-regulated, and 353 in
women of which 268 (76%) were up-regulated.
Probes common to men and women
We next determined whether there was overlap of up-
regulatedanddown-regulatedprobesinmenandwomen.At
FIG. 2. Analysis by ALR in GH-treated men and women. ALR results for all probes on the G4112F microarray (see Materials and Methods). Range
of log2 fold change vs. frequency, range of P value vs. frequency, and relationship between P value and log2 fold change for GH-treated men (A–
C) and women (D–F) are shown. In C and F, the horizontal lines represent threshold P Ͻ 0.05 corresponding to a FDR of 5%, whereas the vertical
lines indicate log2 fold change of 0.585 (equivalent to an absolute fold change of 1.5). The open circles represent the transcripts that show a
greater than 1.5-fold change and FDR less than 5% in response to GH treatment.
FIG. 3. Effect of GH compared to inter-individual variation on probes
expressed above background. The effects of GH treatment relative
to within-individual variation was determined as described in the
Materials and Methods. Variation due to GH within individuals (GH)
was compared with variation between individuals (Interperson). In the
box plots for men and women, solid lines represent the median, the
box represents the middle 50% of data, and whiskers represent 2 SD.
TABLE 1. Probe regulation in GH-treated men and
women.
Regulation
Treatment group
Women Men
No. of probes
Detected (% of all probes) 27412 (63) 27383 (63)
Up (%)a
13557 (49) 12997 (47)
Down (%)a
13855 (51) 14386 (53)
Maximum fold change
Up 2.02 2.09
Down 1.82 1.96
Fold change Ͼ1.5
Up 8 (50) 36 (73)
Down 8 (50) 13 (27)
Unadjusted P Ͻ 0.05
Up (%)a
1465 (52) 1795 (57)
Down (%)a
1334 (48) 1336 (43)
FDR Ͻ0.05
Up (%)a
268 (76) 28 (68)
Down (%)a
85 (24) 13 (32)
ALRs were used to identify differentially expressed probes as described
in Materials and Methods. Results for all probes with an expression
level of 6 on a log2 scale at either baseline or after treatment.
a
Percentage of probes regulated up or down.
4706 Mitchell et al. GH Doping Detection by Blood Microarray J Clin Endocrinol Metab, December 2009, 94(12):4703–4709
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.
5. a FDR less than 0.05, no common probes were identi-
fied. Ranking the top 100 probes in men and women by
fold change, three common up-regulated genes and
three common down-regulated genes were identified.
The three up-regulated genes were IGF-II, MED18, and
PDK4. The three common down-regulated genes were
AREG, ARG1, and CYYR1.
Discriminant probe sets
We applied a multivariate analysis approach using
SDDA to identify groups of probes to discriminate be-
tween the baseline and GH-treated results for men and
women. A group of three probes for men (data not pre-
sented) and one of five probes for women (data not pre-
sented) could discriminate between the two treatment
points but did not have the largest average fold change.
However, these groups of probes performed very poorly
on cross-validation. This approach did not identify groups
of probes that could discriminate between treatment in
men or women.
Validation by Q-PCR
Six differentially expressed genes were selected on the
basis of altered expression due to GH treatment for vali-
dation in men and/or women. An additional gene, CD46,
was selected because it was unaltered by GH treatment in
either men or women. In agreement with the microarray
data,therewerenochangesinCD46expressionbetweenwk
0 and 8 in both men and women (Fig. 4). In women, there
was good agreement between microarray and Q-PCR re-
sults. Assays for HSPC159, ITGB3, OLFM4, and TUBB1
confirmed that these mRNAs were up-regulated after GH
treatment (Fig. 4). We were unable to validate the increase
in expression of APOL6 or TBC1D25 in men. A decrease
in OLFM4 expression was observed; however, this was
lessthanthechangeobservedbymicroarray(Fig.4).Thus,
validationofmicroarrayresultsbyQ-PCRforseventarget
genes was mixed with agreement in women but poor
agreement in men.
Discussion
This is a novel report of the effects of systemic GH treatment
on transcripts in peripheral blood in men and women. GH
treatment unequivocally increased circulating IGF-I levels.
However, expression changes in peripheral blood were
small, with maximal effects approximately 2-fold up-regu-
lation and down-regulation. Approximately 11% of all
probes expressed were significantly regulated with more
differentially regulated in women. The effects of GH in
men and women were different, with only three of the top
100 up-regulated or down-regulated probes ranked by
fold change being common. The effect of GH was of sim-
ilar magnitude to the variation between individuals at
baseline. In short, a supraphysiological dose of GH in-
duced a robust increase in IGF-I, whereas its transcrip-
tional effects are modest and of similar magnitude to the
variation between individuals.
There are potential explanations for the modest tran-
scriptional effects of GH. It may be that GH does not exert
uniform effects on transcription in a mixed cell popula-
tion, despite the evidence that several subpopulations of
peripheral blood cells are GH responsive (9, 10). There
may be much larger effects in particular subpopulations
that are masked by underrepresentation in the total sam-
ple. It is possible that the measured endpoint reflects
the net result of multiple, possibly opposing effects on the
same transcript in different subpopulations. Despite the
heterogeneous cell population, Eady et al. (21) have dem-
onstrated that a proportion of transcripts remain remark-
ably constant with time and that the major variation was
between individuals, a finding in agreement with ours. A
recent pilot study reported that GH treatment induced
GH - Treated Men
-2.0
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
APOL6 CD46 HSPC159 ITGB3 OLFM4 TBC1D25 TUBB1
Gene of Interest
Microarray QPCR
GH - Treated Women
-1.5
-1.0
-0.5
0.0
0.5
1.0
1.5
APOL6 CD46 HSPC159 ITGB3 OLFM4 TBC1D25 TUBB1
Gene of Interest
Microarray QPCR
FIG. 4. Validation of target genes via TaqMan gene expression assay.
For the seven targets identified, expression relative to the
housekeeping gene GAPDH was calculated as described in the
Materials and Methods. Q-PCR results are shown as Ϫ⌬⌬CT (cycle
threshold) for men and women for the difference in expression
between baseline and final treatment vs. ALR, mean Ϯ SE. TaqMan
gene expression assay IDs are as follows: GAPDH, Hs99999905_m1;
CD46 Hs00387246_m1; OLFM4, Hs00197437_m1; APOL6,
Hs00229051_m1; TBC1D25, Hs00412781_m1; HSPC159,
Hs00204379_m1; ITGB3, Hs01001469_m1; and TUBB1,
Hs00258236_m1.
J Clin Endocrinol Metab, December 2009, 94(12):4703–4709 jcem.endojournals.org 4707
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.
6. significant changes in gene expression with maximum of
up to 9-fold in the peripheral blood of children with GH
deficiency but none fulfilling criteria for differential ex-
pression in children with Turner’s syndrome (22). This
observation suggests that the transcriptome response to
GH therapy in a deficiency state is far greater than that in
a replete state. Variation between individuals, coupled
with the weak effect of GH supplementation, indicates
that using whole peripheral blood, rather than individual
cell populations, may confound the ability to detect GH
doping by this method.
The effects of GH on peripheral blood leukocytes were
different in men and women. This was true at the probe
levelandattheleveloftheunderlyingbiologythatwehave
also assessed by gene set enrichment analysis (data not
shown). Women are less GH responsive (6). Even though
women received a larger dose of GH per kilogram body
weight, there was a lesser increase in IGF-I levels (Fig. 1).
Sex steroids can modulate GH action, with estrogens at-
tenuating and androgens enhancing the biological effects
of GH (23, 24). Indeed, there is evidence for a number of
differential effects of sex hormones on immune responses
in humans (25). Thus, it is not unexpected that we have
demonstrated gender-specific effects.
A similar percentage of probes was identified as being
differentially regulated in both men and women (Table 1).
However, considering the results by FDR, more probes
were differentially regulated in women at all cutoffs. This
observation may reflect greater statistical power from 13
vs. seven participants in women and men, respectively,
and may explain the apparently better concordance be-
tween microarray and Q-PCR data for women. From the
sample of genes selected, we identified two false-positive
probes in men and an additional probe with an expres-
sion change less than that predicted by microarray. All
genes selected in women performed as expected. Over-
all, our results are representative of the small effect of
GH on this system relative to the inter-individual vari-
ation in expression.
Our ranking by fold change identified six genes that
were commonly and differentially regulated by GH in men
and women, three of which were up-regulated (IGF-II,
MED18, and PDK4) and three down-regulated (CYYR1,
AREG, and ARG1). These genes are involved in growth,
transcriptional regulation, differentiation, and metabo-
lism. Unlike IGF-I, IGF-II, a potent cellular growth factor,
is not known to be a GH-regulated gene. However, it is
up-regulated by IGF-I in the kidneys of mice (26), sug-
gesting the GH-induced response may have been indirect.
MED18 is a component of the Mediator complex, a co-
activator involved in the regulated transcription of nearly
all RNA polymerase II-dependent genes, suggesting a role
of GH in transcriptional activation (27). Pyruvate dehy-
drogenase kinase isozyme 4 (PDK4), a mitochondrial en-
zyme, inactivates the pyruvate dehydrogenase complex,
the key enzyme converting pyruvate to acetyl-coenzyme A
for oxidation in the tricarboxylic acid cycle. Thus, the
effect is predicted to inhibit glucose use. We recently re-
ported that GH treatment represses the transcription of
PDK4 in muscle (28). This effect is opposite to that ob-
served in leukocytes, suggesting that glucose metabolism
may be regulated by GH in a tissue-specific manner.
Among the down-regulated transcripts, CYYR1 is lo-
cated on human chromosome 21, and its product has no
similarity to any known protein and is of unknown func-
tion (29). Amphiregulin (AREG), a member of the epider-
mal growth factor family promotes the growth of normal
epithelial cells but inhibits the growth of certain carci-
noma cell lines (30). Qi et al. (31) have reported that am-
phiregulin is expressed in the basophil population of pe-
ripheral leukocytes and responds to IL-3 stimulation,
suggesting a role in type 2 immune response. Amphiregu-
lin was up-regulated by GH in peripheral leukocytes of
GH-deficient children in contrast to our finding in adults
(22). The reason is not known but could reflect an age-
dependant effect. Arginase 1 (ARG1) is the enzyme in-
volved in final step of the urea cycle, converting L-arginine
into L-ornithine and urea in the liver. Arginase 1 has been
identified in macrophages, where it converts L-arginine to
nitricoxide,whichenhancestumoricidalactivity(32).The
significance of these transcript changes in peripheral
blood leukocytes is unknown.
We attempted to identify small groups of probes with
diagnostic potential for GH administration in men and
women. A multivariate analysis approach using SDDA
identified probes that could discriminate between baseline
and final treatment samples. However, on cross-valida-
tion including all technical replicates, the discrimination
was not maintained. Further analyses may identify probe
sets that are discriminant over a certain threshold. Al-
though we identified a number of probes that were regu-
lated consistently across all individuals within a treatment
group, most of these probes displayed very limited fold
changes. The weak effect of GH on the transcriptome of
peripheral blood prevents identification of diagnostic
markers of GH treatment.
In summary, the small fold changes in probe expression
levels indicate that GH induces a subtle effect despite caus-
ing unequivocal increases in circulating IGF-I levels. Be-
cause GH-induced alterations in gene expression are small
and of the order of variation between individuals, dis-
criminant gene signatures cannot be defined. We conclude
that gene expression analysis of peripheral blood leuko-
cytes is unlikely to be a viable approach for the detection
4708 Mitchell et al. GH Doping Detection by Blood Microarray J Clin Endocrinol Metab, December 2009, 94(12):4703–4709
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.
7. ofGHdopingandoffersnoadvantagetothemeasurement
of circulating protein markers of GH abuse.
Acknowledgments
We are indebted to our volunteers for their participation in the
study. We thank Jennifer Hansen and Angela Peris for excellent
clinical assistance. We thank Novo Nordisk for the supply of
human GH (Norditropin).
Address all correspondence and requests for reprints to: Pro-
fessorKenHo,PituitaryResearchUnit,GarvanInstituteofMed-
ical Research, 384 Victoria Street, Darlinghurst New South
Wales 2010, Sydney, Australia. E-mail: k.ho@garvan.org.au.
This work was supported by World Anti-Doping Agency.
Disclosure Summary: The authors have nothing to disclose.
References
1. Le Roith D, Bondy C, Yakar S, Liu JL, Butler A 2001 The somato-
medin hypothesis: 2001. Endocr Rev 22:53–74
2. World Anti-Doping Agency 2009 World Anti-Doping Code 2009.
http://www.wada-ama.org/rtecontent/document/code_v2009_En.pdf.
Montreal: World Anti-Doping Agency
3. World Anti-Doping Agency 2009 The 2009 prohibited list. http://
www.wada-ama.org/rtecontent/document/PL2009_EN_Final.pdf.
Montreal: World Anti-Doping Agency
4. Nelson AE, Ho KK 2008 A robust test for growth hormone doping:
present status and future prospects. Asian J Androl 10:416–425
5. Holt RI, So¨nksen PH 2008 Growth hormone, IGF-I and insulin and
their abuse in sport. Br J Pharmacol 154:542–556
6. NelsonAE,MeinhardtU,HansenJL,WalkerIH,StoneG,HoweCJ,
Leung KC, Seibel MJ, Baxter RC, Handelsman DJ, Kazlauskas R,
Ho KK 2008 Pharmacodynamics of growth hormone abuse biomar-
kers and the influence of gender and testosterone: a randomized
double-blind placebo-controlled study in young recreational ath-
letes. J Clin Endocrinol Metab 93:2213–2222
7. Bidlingmaier M, Suhr J, Ernst A, Wu Z, Keller A, Strasburger CJ,
Bergmann A 2009 High-sensitivity chemiluminescence immunoas-
says for detection of growth hormone doping in sports. Clin Chem
55:445–453
8. NguyenTV,NelsonAE,HoweCJ,SeibelMJ,BaxterRC,Handelsman
DJ, Kazlauskas R, Ho KK 2008 Within-subject variability and analytic
imprecision of insulinlike growth factor axis and collagen markers:
implications for clinical diagnosis and doping tests. Clin Chem 54:
1268–1276
9. Clark R 1997 The somatogenic hormones and insulin-like growth
factor-1: stimulators of lymphopoiesis and immune function.
Endocr Rev 18:157–179
10. Redelman D, Welniak LA, Taub D, Murphy WJ 2008 Neuroendo-
crine hormones such as growth hormone and prolactin are integral
members of the immunological cytokine network. Cell Immunol
252:111–121
11. Shimizu K, Adachi K, Teramoto A 2005 Growth hormone enhances
natural killer cell activity against glioma. J Nippon Med Sch 72:
335–340
12. Chen HT, Schuler LA, Schultz RD 1998 Growth hormone receptor
and regulation of gene expression in fetal lymphoid cells. Mol Cell
Endocrinol 137:21–29
13. Baxter RC, Brown AS, Turtle JR 1982 Radioimmunoassay for so-
matomedin C: comparison with radioreceptor assay in patients with
growth-hormone disorders, hypothyroidism, and renal failure. Clin
Chem 28:488–495
14. Smyth GK 2004 Linear models and empirical Bayes methods for
assessing differential expression in microarray experiments. Stat
Appl Genet Mol Biol 3:Article3
15. Ritchie ME, Diyagama D, Neilson J, van Laar R, Dobrovic A,
Holloway A, Smyth GK 2006 Empirical array quality weights in the
analysis of microarray data. BMC Bioinformatics 7:261
16. Ihaka R, Gentleman R 1996 R: a language for data analysis and
graphics. J Comput Graph Stat 5:299–314
17. Gentleman RC, Carey VJ, Bates DM, Bolstad B, Dettling M, Dudoit
S, Ellis B, Gautier L, Ge Y, Gentry J, Hornik K, Hothorn T, Huber
W, Iacus S, Irizarry R, Leisch F, Li C, Maechler M, Rossini AJ,
Sawitzki G, Smith C, Smyth G, Tierney L, Yang JY, Zhang J 2004
Bioconductor: open software development for computational biol-
ogy and bioinformatics. Genome Biol 5:R80
18. Dudoit S, Fridlyand J, Speed TP 2000 Comparison of discrimination
methods for the classification of tumors using gene expression data.
Tech Rep 576. Berkeley, CA: University of California, Berkeley
19. Storey JD, Tibshirani R 2003 Statistical significance for genome-
wide studies. Proc Natl Acad Sci USA 100:9440–9445
20. Livak KJ, Schmittgen TD 2001 Analysis of relative gene expression
data using real-time quantitative PCR and the 2Ϫ⌬⌬CT
method.
Methods 25:402–408
21. Eady JJ, Wortley GM, Wormstone YM, Hughes JC, Astley SB, Foxall
RJ, Doleman JF, Elliott RM 2005 Variation in gene expression profiles
ofperipheralbloodmononuclearcellsfromhealthyvolunteers.Physiol
Genomics 22:402–411
22. Whatmore AJ, Patel L, Clayton PE 2009 A pilot study to evaluate gene
expression profiles in peripheral blood mononuclear cells (PBMCs)
from children with GH deficiency and Turner syndrome in response to
GH treatment. Clin Endocrinol (Oxf) 70:429–434
23. Gibney J, Wolthers T, Johannsson G, Umpleby AM, Ho KK 2005
Growth hormone and testosterone interact positively to enhance
protein and energy metabolism in hypopituitary men. Am J Physiol
Endocrinol Metab 289:E266–E271
24. Leung KC, Doyle N, Ballesteros M, Sjogren K, Watts CK, Low TH,
Leong GM, Ross RJ, Ho KK 2003 Estrogen inhibits GH signaling by
suppressing GH-induced JAK2 phosphorylation, an effect mediated
by SOCS-2. Proc Natl Acad Sci USA 100:1016–1021
25. Bouman A, Heineman MJ, Faas MM 2005 Sex hormones and the
immune response in humans. Hum Reprod Update 11:411–423
26. Svensson J, Tivesten A, Sjo¨gren K, Isaksson O, Bergstro¨m G, Mohan
S, Mo¨lne J, Isgaard J, Ohlsson C 2007 Liver-derived IGF-I regulates
kidney size, sodium reabsorption, and renal IGF-II expression.
J Endocrinol 193:359–366
27. Lariviere L, Geiger S, Hoeppner S, Rother S, Strasser K, Cramer P
2006 Structure and TBP binding of the Mediator head subcomplex
Med8-Med18-Med20. Nat Struct Mol Biol 13:895–901
28. Sjo¨gren K, Leung KC, Kaplan W, Gardiner-Garden M, Gibney J, Ho
KK 2007 Growth hormone regulation of metabolic gene expression
in muscle: a microarray study in hypopituitary men. Am J Physiol
Endocrinol Metab 293:E364–E371
29. Vitale L, Casadei R, Canaider S, Lenzi L, Strippoli P, D’Addabbo P,
Giannone S, Carinci P, Zannotti M 2002 Cysteine and tyrosine-rich
1 (CYYR1), a novel unpredicted gene on human chromosome 21
(21q21.2), encodes a cysteine and tyrosine-rich protein and defines
a new family of highly conserved vertebrate-specific genes. Gene
290:141–151
30. Xian CJ 2007 Roles of epidermal growth factor family in the reg-
ulation of postnatal somatic growth. Endocr Rev 28:284–296
31. Qi Y, Operario DJ, Oberholzer CM, Kobie JJ, Mosmann TR 2009
The expression and regulation of amphiregulin on human basophils
and CD4 T cells. J Immunol 182:36.16 (Abstract)
32. Chang CI, Liao JC, Kuo L 2001 Macrophage arginase promotes
tumor cell growth and suppresses nitric oxide-mediated tumor
cytotoxicity. Cancer Res 61:1100–1106
J Clin Endocrinol Metab, December 2009, 94(12):4703–4709 jcem.endojournals.org 4709
The Endocrine Society. Downloaded from press.endocrine.org by [${individualUser.displayName}] on 29 April 2015. at 20:16 For personal use only. No other uses without permission. . All rights reserved.